From first principles, upward.

A collaborative community of engineers and researchers learning, building, and studying AI/ML through first-principles understanding, hands-on implementation, experimentation, research, and open-source collaboration.

Machine Learning Curriculum · Learning Roadmap · Contribute

Our work spans the AI/ML stack — from mathematical foundations to production AI systems.

• Machine Learning — supervised, unsupervised, semi-supervised, and self-supervised learning
• Neural Networks & Deep Learning — optimization, training dynamics, architectures, representation learning
• Natural Language Processing — classical NLP, embeddings, transformers, language models
• Computer Vision — image learning, CNNs, vision transformers, multimodal systems
• Reinforcement Learning — value-based methods, policy learning, deep RL
• Generative AI & LLMs — pretraining, adaptation, prompting, retrieval, RAG, inference
• Agentic AI — tool use, planning, memory, orchestration, and evaluation
• AI Evaluation & Reliability — metrics, benchmark design, robustness, safety, and continuous evaluation
• ML Engineering & MLOps — experimentation, serving, monitoring, pipelines, and production reliability
• AI Systems & Infrastructure — training systems, inference systems, retrieval infrastructure, observability, and AI platforms
• Research & Reproduction — papers, ablations, experiments, and deeper technical investigations

An open, contributor-built, executable curriculum for learning AI/ML from first principles to advanced systems.

It is designed to work in both directions:

Learner → Contributor
Pick a topic, learn it deeply, implement it, experiment with it, and submit your work.

Contributor → Future Learner
Accepted contributions become reusable learning material for the next person.

Our learning contract is:

LEARN → DERIVE → BUILD → USE → EXPERIMENT → REFLECT

The goal is not another repository full of disconnected notebooks. The goal is a coherent, reviewable, executable textbook built by people learning through contribution.

1. Learn deeply, not superficially. Understanding should survive beyond a library call or copied notebook.
2. Build what you learn. Implementation exposes gaps that passive reading hides.
3. Experiment instead of hand-waving. Claims should be tested where practical.
4. Document failures, assumptions, and trade-offs. Knowing when something breaks is part of knowing how it works.
5. Make individual contributions visible. Work happens through branches, commits, issues, reviews, and pull requests.
6. Teach the next learner. A contribution is strongest when someone else can learn from it.
7. Prefer durable knowledge over hype. New techniques matter when they improve the learning graph — not simply because they are fashionable.

Learn
  ↓
Build
  ↓
Experiment
  ↓
Contribute
  ↓
Review
  ↓
Teach the next learner
  ↓
Ascend together

The Syndicate is built around contribution rather than passive membership.

If a topic interests you:

1. find or propose a learning unit;
2. understand its prerequisites;
3. study, derive, implement, and experiment;
4. document what you learned;
5. open a pull request;
6. improve it through peer review;
7. leave behind something useful for the next learner.

Start with the Machine Learning Curriculum.

From first principles, upward.
_{Learn deeply. Build rigorously. Ascend together.}